Colon Cancer Risk Susceptibility Technique to Be Identified Through Genome-wide Screen

---

Individuals at risk for colon cancer can be identified early by measuring inherited genetic risk factors. Currently technologies are being used to measure vast amounts of genetic variation to identify such predictors with the goal of developing public health strategies focused on early detection. This work has been funded by Genome Canada through the Ontario Genomics Institute under the Applied Genomics and Proteomics Research in Human Health Competition and has been branded the Assessment of Risk for Colon Tumours in Canada (ARCTIC) Project.

Colorectal cancer a preventable public health burden
Colorectal cancer is a malignancy of the large bowel. It is the third most common cancer and the third most common cause of death from cancer in Canada for both men and women. According to the Canadian Cancer Society, in 2006 alone, there will be 20,000 new cases and 8,500 deaths from colon cancer with an age standardized incidence rate of 63/100,000. This disease accounts for 12.4 percent of all cancer deaths and an estimated 120,000 years of life will be lost in 2006. Colorectal cancer ranks as the second most frequent cause of cancer deaths among Canadians with one in 18 persons developing colon cancer at some time in their life {4369}.

Like so many human malignancies, early detection of colorectal cancer is key to survival. Since early disease generally goes undetected due to early tolerance of intracolonic mass lesions, detection at a curable stage occurs infrequently and population-wide screening programs are not yet in place or poorly accepted.

Approximately 20 percent of 'sporadic' cases of colorectal cancer have a familial component. This suggests that pre-morbid identification of individuals at risk could be possible through genetic testing. Individuals at risk could enroll in intensive screening programs, or potentially in chemoprophylaxis studies, which would be designed to reduce risk through the chronic administration of pharmaceutical agents. Matching small variations in the human genome has given way to the objective identification of human genetic variants associated with a disease through whole genome scanning techniques. These techniques are ideally suited to examining human colon cancer but are predicated on an accurate map of the sites of variability.

International researchers create map of human genetic variability
To provide a framework for the discovery of genetic determinants of disease, such as colorectal carcinoma, three international consortia have worked over the last decade to map the sites and frequency of human natural genetic variation. The Human Genome Project, which was completed in 2001 used the combined efforts of publicly and privately-funded research to reveal the sequence 99 percent of the entire human genome {3563}.

Following this milestone, the SNP Consortium, which examined individuals, revealed wide-spread sites of single nucleotide sequence variability, deposited into the newly created dbSNP now containing over 12 million sites. In recognition of remarkable regions of linkage disequlibrium, genome-wide mapping of human haplotypes was launched in 2002 by an international consortium, the International HapMap Consortium, to provide a cost-effective way of hunting for genomic diversity though the identification of sites in high linkage disequilibrium. In addition, the use of SNP panels provide efficient and comprehensive analyses of individuals tested in the context of disease association studies.

ARCTIC Project
The ARCTIC project, conducted by McGill University and Cancer Care Ontario researchers, was funded in 2004 by Genome Canada through the Ontario Genomics Institute, the Ontario Cancer Research Network (the predecessor to the Ontario Institute of Cancer Research) and Genome Quebec to perform high throughput genotyping of DNA from colon cancer patients and controls to discover genetic predictors of disease risk.

The ARCTIC project was based on the hypothesis that individual genetic differences may affect cancer risk. The coincidence of such functionally relevant variability may cause significant risk for an individual and collectively account for a large proportion of population attributable risk. Changes that would individually confer only minor increased disease susceptibility could collectively cause significant individual risk. Finding these genetic determinants would allow researchers to predict individual risk and customized health promotion programs could be designed, leading the way in this new era of personalized medicine.

To execute this $10 million program, a multidisciplinary team was assembled using resources and infrastructure found throughout Ontario and Quebec.

Available tissue and genotyping resources
The ARCTIC project was possible because of access to the American National Institutes of Health Familial Colon Cancer Registry tissue collection. This program, initiated in 1998, coordinates five collection centres, funding the collection of familial and sporadic colorectal cancer tumours and peripheral blood. The FCCR program uses a Principle Investigator Steering Committee which oversees access, acting as a peer review committee. The Ontario Familial Colon Cancer Registry, with more than 1,500 cases and carefully matched population-based controls, was made available for this project in order to measure potential cancer-associated genotypes.

Genotyping of all study material, including primary study sets and all validation cases was performed at the McGill University and Genome Quebec Innovation Center. This facility, with state-of-the-art instrumentation including Affymetrix expression and genotyping platforms, Illumina Golden Gate and Infinium genotyping technologies and lower volume technologies such as qPCR and Genome Lab’s SNPstream, contributed to the International HapMap project by testing more than 150,000 STSs on chromosomes 2 and 4p {4298}. This facility performs contract-based genotyping, expression and DNA sequencing for both academic and industrial researchers and remains critical to Montreal's innovation infrastructure.

Statistical analysis uses facilities in Toronto and London
High throughput genomic techniques produce unprecedented amounts of data requiring intensive computing, novel mathematics and analytical resources. The ARCTIC project has harnessed the resources of three institutes, including the Hospital for Sick Children in Toronto, the University of Western Ontario (London) and the University of Toronto. The core statistical group has identified genetic associations through univariate testing and is engaged in leading edge development of multivariate predictive models of disease risk, using a combination of classic epidemiologic risk factors and inherited genetic diversity. Once validated on internal and external data, the final model will form the analytic basis of the test and will be embedded in a genetic test package.

Execution of the experimental plan
Using large-scale techniques, measurements were taken of inherited genomic diversity in patients with colon cancer and matched controls. Recent advances in massively parallel techniques allowed whole genome scans of human genomic diversity. Using Affymetrix GeneChip genotyping arrays, measurements showed 600,000 distributed polymorphisms and 10,000 non synonymous coding SNPs in each of 1,200 Ontario colorectal cancer patients and 1,200 controls. The resulting dataset, measuring almost 1.5 billion elements, was analysed using the Chi Square statistic with significance thresholds determined by generating and analysing 10,000 random permutations of the data.

The measurement of huge numbers of variables in any association study will identify many false, but statistically significant, differences. This problem was approached through the analysis of only highly significant polymorphic sites derived from the primary screen in independently derived cohorts of patients. Based on this approach, 1,100 polymorphisms were selected for validation in a population of 1,200 sporadic cases and 1,200 controls collected by the Newfoundland Familial Colon Cancer Registry (NFCCR), a registry patterned after the NIH-sponsored sites and the Seattle Familial Colon Caner Registry (SFCCR). Of approximately 100 significant discriminators of disease, 20 will be further validated in an additional 5,000 patients and an equal number of controls from Glasgow, Scotland and Haifa, Israel.

Prospective validation uses PLCO cohort
The ultimate measure of the usefulness of the set of polymorphisms will come from prospective validation and the generation of test performance characteristics such as positive predictive value and negative predictive value. To perform such an evaluation in a truly prospective fashion would require decades of follow-up observation. To make the research clinically valuable as soon as possible, a partnership was formed with an existing large research effort, the NIH-sponsored American Prostate, Lung, Colon, Ovary (PLCO) cohort, which has DNA samples and data collected since 1993. Ultimately the usefulness of this individual risk stratification tool will be demonstrated through its integration into programs of colon cancer screening and care, with detailed health economic evaluation.

A view of the future
Our predictors of disease, while suggesting interesting biological hypotheses, also provide the substrate for cancer control through early assessment of risk through genetic testing. Moreover, the ARCTIC data base provides a rich resource for the evaluation of other sources of genetic variation such as copy number variability which may also help determine colon cancer disease risk.

Genetic testing for disease susceptibility has historically followed existing demand rather than anticipating demand or creating new initiatives. The need for diagnostic tests often comes long before such tests are readily available. For example, demand by clinicians for tests to identify individuals at risk for diseases such as cystic fibrosis, adenomatous polyposis coli (APC), hereditary non-polyposis colorectal cancer, retinoblastoma and hereditary breast cancer preceded their availability by decades. Development and marketing efforts for these tests have focused on a simple ready market, being offered largely to identifying individuals at risk with little or no manufacturer-lead research or market development.

Tests, such as the ARCTIC Test, have obvious applications but, like proprietary drugs, will require field testing, clinical validation and a careful health economic analysis conducted during screening trials of the new technology. This clinical development program will result in a high value tool, fostered by funding from federal and provincial governments and the willingness of Canadian health care agencies to help demonstrate its clinical usefulness.
Genome Canada has recently announced calls both for a competition focused on new technology development for genomics and proteomics and for input from the scientific community in Canada on what strategic initiatives would be best served by genomics and proteomics research and resources, drawing on the expertise built up over the past five years, and addressing issues, problems and opportunities of national importance.